In Hwa Jeong, Heeyoung Kim, Hyunyong Hwang
Competency-based medical education emphasizes formative assessment and longitudinal portfolio review; however, the resulting workload is increasingly difficult for faculty members to sustain. Generative artificial intelligence (GenAI) and large language models (LLMs), such as ChatGPT, Gemini, and Claude, offer opportunities to alleviate these burdens while enriching teaching and learning. This narrative review synthesizes peer-reviewed literature published between January 2023 and April 2026, retrieved from PubMed/MEDLINE, Scopus, Web of Science, Google Scholar, and Education Resources Information Center (ERIC), on the practical applications of GenAI and LLMs in medical education. Five core domains were identified: simulation of clinical reasoning, formative assessment and e-portfolio evaluation, support for self-directed learning, curriculum design, and communication skills training. Important challenges and ethical considerations are also discussed, including cognitive offloading and the potential erosion of critical thinking, hallucinations and threats to academic integrity, data privacy and institutional compliance concerns, and the risk of widening the digital divide. Future directions include multimodal LLMs and integrated clinical simulations, longitudinal competency tracking, faculty development and institutionalization of artificial intelligence (AI) literacy, standardized evaluation frameworks and regulatory guidance, and ambient AI in smart educational environments. When integrated thoughtfully, GenAI and LLMs may reduce faculty workload and enrich the learner experience while preserving educators’ authority over competency judgments and the humanistic core of medical education. Realizing this potential will require coordinated investment in faculty’s AI literacy, robust governance, equity-focused deployment, and rigorous validation of AI-enhanced educational tools.